Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

WIP: Release/1.3.0 #365

Merged
merged 167 commits into from
Feb 7, 2020
Merged

WIP: Release/1.3.0 #365

merged 167 commits into from
Feb 7, 2020

Conversation

mpu-creare
Copy link
Contributor

@mpu-creare mpu-creare commented Feb 5, 2020

This is waiting on #364 and #361 . But starting so that I can make the release notes.

mpu-creare and others added 30 commits November 6, 2019 15:52
Changes "create_data_array" to "create_dataarray" to be consistent with xarray naming convention in methods.

This feature adds the ability to symmetrically go between UnitsDataArray -> netcdf -> UnitsDataArray.  The goal is move away from pickle in the lambda function interface and more generally, provide users a way to share binary outputs and get them back in PODPAC.
Lambda node by default uses "invoke" instead of S3. Includes handler improvements to consolidate the "pipeline" data model that is passed amongst the various triggers.
Data is passed back and forth via netcdf.

API Gateway and S3 triggers still needs testing.

In the future, I think we should deprecate the "eval_s3" and "eval_api" methods since these are no longer relevant. We should keep the ability to add extra triggers for S3 and APIGateway since there may be other use cases that need this, but the PODPAC only evaluation will be done through invoke.
…vor of classmethods `UnitsDataArray.create()` and `UnitsDataArray.open()`

This is much more clear to me and completes discussion in #334
…hould be overwritten by the aws-improvements branch
Update pyproj to 2.4 in the process. Note the proj4 string is no longer the best way to communicate string based coordinate reference systems. Internally, we should use "wkt2" based on https://proj.org/faq.html#what-is-the-best-format-for-describing-coordinate-reference-systems
…tive_coordinates as an attr so it can be evaluated from pipeline definitions
The `queryStringParameters` are already a dict, so they don't need parsed with `urllib`.

This also merges any settings passed in the event with the default pipeline, then again with the settings on the AWS instance.
… completely consistent with the test, at least not for pyproj 2.4.
…en for a while. I'm now catching the native_coordinates definition at instantiation to make sure it's a Coordinates object. It would be and OrderedDict if created throught Node.from_definition...

Also, native_coordinates cannot be None for an Array datasource, so had to update the tests.
…nd is corrected in the aws-improvements branch
mpu-creare and others added 15 commits February 5, 2020 14:34
…inates can be created from multiple of pi/2, just not theta = 0 which throws the UndefinedRotationError. All multiples of pi/2 should be Uniform coordinates instead. Fixed.
* There was a recursion error due to validation of outputs on the compositor
* Missed adding the 'crs' to the UniformCoordinates case of from_geotransform in coordinates
* Added a "crs" attribute to the Rasterio node. For some reason, the terrain tiles file no longer contain information about the CRS?!? THis is strange and new.
* added intake to list of dependencies.
* setting default CRS in terrain tiles as a fallback in case the CRS is not in the file
…with wkts at times, see notebooks/ams-short-course/03-data-access-harmonization-processing.ipynb .. i.e. TerrainTiles with bilinear interpolation.
…d attributes should help a lot with this.

* A big issue was the cache lock being active while looking for a cached attribute, and then while computing the hash another cache looking is requested, locking up the whole works, that's why I had to explicitly set self.outputs (it was recursive with self.sources)
* The validation has a circular reference between self.source_coordinates and self.sources
* This is a strange one, I think it's a change in how zarr's indexing works -- it no longer accepts numpy arrays
* This transforms any numpy array or list to a slice.
@mpu-creare
Copy link
Contributor Author

A number of these bug fixed really need some unittests!

@mpu-creare mpu-creare merged commit cd5ed38 into master Feb 7, 2020
@mpu-creare mpu-creare deleted the release/1.3.0 branch February 7, 2020 20:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants